SSD Write Endurance 25nm Vs 34nm

[Hopalong X]

It is rated at 1 year after hitting MWI-1.

How far was the M4 past the MWI-1?

It hit MWI-1 at 178TiB. It was up to around 600TiB of writes if memory serves.

Way past MWI 1. If it had been stopped at that point it should have held its data as speced.
After that holding data is questionable to length of time.
Same as asking how long it will continue to write data.

You are all making the new specs with the testing.

[bluestang]

May be we should take a break on Vertex Turbo and have it sit idle, powered off for few days to see if it still keeps going! I have a vague memory that bluestang has been powering it off during the day on working days.

Nope, I don't power it off. It's still connected. I just usually have to boot into XP during the day, except for this past week I've been staying in Win7 to try and catch the M4

[B.A.T]

This was my last result update.
m4
768.5687 TiB
2749 hours
Avg speed 88.43 MiB/s.
AD gone from 171 to 169.
P/E 13330.
Value 01 (raw read error rate) has changed from 43 to 46.
MD5 OK.
Reallocated sectors : 00

The M4 passed MWI-1 at 170.8 TiB.

There were no problems before I turned my rig off the 28th oct. It just didn't like 9 days without power.

[sergiu]

Sad... I was really hoping to see M4 past 1TB. Maybe P/E rating is not conservative at all considering the requirements. I believe it would be better to add a 7-14 days of rest between each 100TB milestone, to check data retention.

[Anvil]

It's not like it was expected to fail but we are way beyond the specs
The spec is 72TB for the Crucial iirc.

I don't think it matters if the pause was 7 or 9 days, it might fail withing hours of disconnecting but we will never know what really happened to B.A.T's drive. It would have been interesting to know if it was a NAND failure or some other part that caused the failure.

As a result of this failure I'll be playing it a bit safe on my drives, I'll start off with 12 hours and then increase in 12hour steps up to 4-5 days.
My drives are turning 500TiB within a few weeks, the Force 3 is not down to MWI 0 yet though, it will be before it hits 500TiB.

[bulanula]

Too bad to see the M4 die like that. BUT where is the C300. That should be able to do 1 PB I think quite easily. Vapor, where are you LOL ???

[Christopher]

So it looks as though even though the M4 had no bad blocks, the NAND just couldn't hold a charge in each cell without power. If that is the case, you'd expect the Vertex Turbo to go much longer while retaining data. I guess at some point these drives just become volatile storage, but I'm more than a little surprised that the M4 never flagged any of these blocks as anything but good.


I still don't know why Crystal Disk Info keeps emailing me with [601] and [701] BAD messages, nor why Windows thinks the Mushkin is dying too. Perhaps they know something I don't?

[bluestang]

Value 01 (raw read error rate) has changed from 43 to 46.

My raw read error rate has actually been decreasing since the start.

I still don't know why Crystal Disk Info keeps emailing me with [601] and [701] BAD messages, nor why Windows thinks the Mushkin is dying too. Perhaps they know something I don't?

Have you looked in your Windows Event Viewer for any errors for that disk?

[Anvil]

You should be able to disable the emails, the CDI messages are most likely based on SMART, there is no other source of info for the drives.
(unless there are issues related to the drive that are found in the Event log, as in e.g. timeouts, which is not very likely as it should show in other ways)

[Christopher]

Of course I can disable the emails, but nothing has changed with the Mushkin in weeks, really. Looking at the SMART data, I can't determine why it is that Windows and CDI seem to think there is something wrong.

The Event Viewer does show a couple of generic SMART based "your drive is going to die" errors. I was just wondering why - if it's based on MWI, then why didn't Windows protest much earlier? It's been at MWI 10 for almost 200TB, and that's the only questionable SMART attribute. I had set up the CDI emails to alert me if anything interesting had happened while I was away, and I'm more curious as to why Windows has a problem now, and not earlier if it's just based on MWI depletion.

[Anvil]

Kingston SSDNow 40GB (X25-V)

455.80TB Host writes
Reallocated sectors : 12

MD5 OK

34.10MiB/s on avg (~47 hours)

--

Corsair Force 3 120GB

01 90/50 (Raw read error rate)
05 2 (Retired Block count)
B1 64 (Wear range delta)
E6 100 (Life curve status)
E7 16 (SSD Life left)
E9 317600 (Raw writes)
F1 422741 (Host writes)

MD5 OK

106.85MiB/s on avg (~47 hours)

power on hours : 1240

B1 is down from 66 to 64.

@Christopher

It would have helped if you posted some screenshots (I know they don't work)
CDI is not some magical tool, it reports when values are below thresholds, some SMART values are said to predict failures and some are not. If the SMART value is said to be predicting failure it's likely that CDI will generate an event/warning.

The SMART attributes that predict failure are the following (for a typical SF2xxx series drive)

1, 5, C4, E6 and E7

[Christopher]

Mushkin Chronos Deluxe 60 Update, Day 46
3.3.2FW

05 5
Retired Block Count

B1 10
Wear Range Delta

F1 453641
Host Writes

E9 350000
NAND Writes

E6 100
Life Curve

E7 10
Life Left

Average 132.11MB/s Avg
Intel RST drivers, Celeron G530 Biostar TH67+

1066 Hours Work
Time 1 month 14 days 10 hours

12 GiB Minimum Free Space

-----------------------------------------------------------

I feel like I'm really making some progress now. Host writes just hit 350TB, and speed seems to be marginally higher.

I traded out another drive for my FW0001 M4 64GB drive last weekend, and I've been playing with it. The lack of SMART attribs to capture host writes is irritating though.

[Anvil]

Kingston SSDNow 40GB (X25-V)

457.38TB Host writes
Reallocated sectors : 12

MD5 OK

33.79MiB/s on avg (~61 hours)

--

Corsair Force 3 120GB

01 94/50 (Raw read error rate)
05 2 (Retired Block count)
B1 63 (Wear range delta)
E6 100 (Life curve status)
E7 14 (SSD Life left)
E9 321580 (Raw writes)
F1 428041 (Host writes)

MD5 OK

106.87MiB/s on avg (~47 hours)

power on hours : 1255

B1 is down from 64 to 63.

[bluestang]

M225->Vertex Turbo 64GB Update:

693.57 TiB (762.58 TB) total
1816.79 hrs (Torture), 2615 hrs (Power-On)
13716 Raw Wear
122.49 MB/s avg for the last 69.19 hours (on W7 x64)
MD5 OK
C4-Erase Failure Block Count (Realloc Sectors) from 11 to 12.
(1=Bnk 6/Blk 2406; 2=Bnk 3/Blk 3925; 3=Bnk 0/Blk 1766; 4=Bnk 0/Blk 829; 5=Bnk 4/Blk 3191; 6=Bnk 7/Blk 937; 7=Bnk 7/Blk 1980; 8=Bnk 7/Blk 442; 9=Bnk 7/Blk 700; 10=Bnk 2/Blk 1066; 11=Bnk 7/Blck 85; 12=Bnk 4/Blk 3192)

I have no plans on doing a power-off state on the M225 until I reach 1 PB

P.S. -- Any mods reading this thread want to get the darn attachments fixed already!

[devsk]

I have no plans on doing a power-off state on the M225 until I reach 1 PB

That's a good plan... I want to see it hit 1PB as well.

[Ao1]

You guys all missed a key milestone. This thread has passed ½ million views

[Ao1]

That's a good plan... I want to see it hit 1PB as well.

Agreed.

This data retention issue does however bring an element of frustration to the testing. Is a year unreasonable? How much of an impact does that requirement have on how the MWI is calculated? At what point should you stop writing and for how long should you then wait?

On one hand a year sounds way to long, but on the other hand I’ve got loads of HDD’s that have been sitting around for years with data backups on them. I’d be pretty hacked if I plugged one of them in to find out it had lost the data. Guess HDD’s will be around for a bit yet, even if they end up only being used for archives.

[Anvil]

Data retention would have shown it's ugly face somewhere down the road anyways.
I'd say it was about time we had to acknowledge that it is a real issue, it is not an issue for normal wear though.

It leaves a lot of questions though.

And now for something completely different

The Samsung 830 series does carry a Host Write attribute.

F1 shows host writes in sectors. (512 byte sectors)



Could be that there are more interesting attributes but F1 and BE are the ones that have changed during testing.

[B.A.T]

Todays update:
Kingston V+100
196.4984 TiB
776 hours
Avg speed 75.40 MiB/s
AD gone from 4 to 1.
P/E?
MD5 OK.
Reallocated sectors : 00



Intel X25-M G1 80GB
20,8113 TiB Host writes
18739 hours
Reallocated sectors : 00
MWI=75 to 65
MD5 OK
45.11 MiB/s on avg

[Anvil]

I'm already missing the m4

[Anvil]

Kingston SSDNow 40GB (X25-V)

458.60TB Host writes
Reallocated sectors : 12

MD5 OK

33.63MiB/s on avg (~72 hours)

--

Corsair Force 3 120GB

01 86/50 (Raw read error rate)
05 2 (Retired Block count)
B1 62 (Wear range delta)
E6 100 (Life curve status)
E7 14 (SSD Life left)
E9 324624 (Raw writes)
F1 432090 (Host writes)

MD5 OK

106.89MiB/s on avg (~72 hours)

power on hours : 1265

B1 is down from 63 to 62.

[bulanula]

Strange thing no reallocated sectors were reported before it just died. Anyone else thinking controller failure and NAND is still OK. Too bad we don't have the equipment to confirm that the NAND is fine etc.

[Christopher]

Maybe something kept the M4 controller from reinitializing on power on? I find it hard to believe that the M4 just quit working after being powered off for a few days. But the drive can't be seen by the bios at all, so its pretty much dead.

That's not satisfying at all.

[johnw]

I'm already missing the m4

I'm thinking that a good way to test in the future is a one-week-on, one-week-off sort of thing (or maybe two or three on, one off). It is interesting to know how much can be written if the SSD is kept powered on all the time, but I think a more useful thing to know is how much can be written while still maintaining a one-week power-off data-retention time. Of course, it would be better to know the value for something like a 3-month power-off data-retention time, but that would take too long to measure if you had to stop writing for 3 months after every 25TiB (or whatever accuracy you want for the measurement).

[johnw]

Maybe something kept the M4 controller from reinitializing on power on? I find it hard to believe that the M4 just quit working after being powered off for a few days. But the drive can't be seen by the bios at all, so its pretty much dead.

I think it is obvious what happened. Same as happened to the Samsung 470. Once the unpowered data-retention time gets down to a few days, powering off the SSD for several days causes most of the flash to lose its charge. Since the SSD stores critical metadata on the flash, the SSD cannot operate once most of the flash has lost its charge.